sub 8
General PA repairs
Replacement 2S power transistors are a bit hard to find and expensive. The SK, NTE replacement are also a bit spendy and any old s**t they can find. I have been studying transistor curves and choosing replacement transistors for years and my latest general replacements are
\MJ21196G NPN and MJ21195G PNP
You might like MJ21194G NPN and MJ121193G PNP also.
They are 250 volt, 16 amp 250 watt and the G is a TO3 form. They are available in other packages with different suffix letters.
Both pairs have identical ratings, but slightly different curves. They are available from Mouse and others at low cost.
For drivers I am using MJE15032G NPN & MJE15033G PNP. They are250volts amp 50 watt TO 220 style 1 (BCE)
If I replace any power transistors on a heat sink I replace all old part numbers with new complementary pairs. Do NOT mix different transistors between different complimentary pairs.
Repairs on one Setton RS-660 PA
I spent 3 days fighting with the right channel PA from my new Setton RS 660. Someone had been here first, always a bad sign. There was a burned off trace from R23 / 24 to L1, R15 was blown and a number of lifted traces from too much heat while someone was playing around. The original Q2 & Q3 2SD381 had been replaced with 2SC2073 and Q4 2SB356 (complement to 381) was replaced with 2SA1930 not complementary pair with 2SC2073. The 100 ohm bias pot RV1 was flaky and around 260 ohms. I got a great education about forward reference diodes (D2) as it was swapped with RV1 on the board layout compared to the schematic.
BTW D2 is a STV-4H and is 4 diodes in series in a single package. The STV-4H is speced 2.10V to 2.36V at 7ma. In a pinch I would guess using 4 silicon diodes in series would work, but the bias might drift a little more.
The PA was removed to the bench and powered with lab power supplies with adjustable current limits to keep it from letting the magic smoke out while trouble shooting. I have no excuse for not using a scope off the bat to find the darn thing had a high frequency oscillation. Hind sight is always 20 / 20. If you do not have a supply with adjustable current limit try a 100W bulb in series with one of the 50V, put ½ or 1A fuses in series with both 50V leave and use jumper the amp power out to the board. This is not a place for a nuwbie to go on an adventure! Tonto say UGH much magic smoke and fried fingers.
When the PA was returned to service in the amp, 1A fuses were add in series with + and – 50V in case of problems. They will become permanent and changed to 3A (200 watt 70% efficiency) after a month or so of driving some 12” or15” CVs.
I replaced Q2 & 3 with MJE15032G and Q4 with the complementary MJE15033G, but oscillations remained. C4 15Pf Q2 base to collector was removed and measured 17Pf. C19 & C20 were also in spec. Increasing value of the 3 caps had no effect on oscillation. The components on the board in the signal path were verified. The Hfe of the MJEs are in the ball park with the original. I could find no curves for the original transistors to compare.
Output transistors were changed from original 2Sxxxx to SK & NTE subs and to MJ21196G NPN and MJ21195G PNP which were left in place. The solder around several TO3 socket connections looked a little gray so all TO3 socket pins were resoldered. An interesting exercise, but no banana.
The oscillation was stopped by adding a .005 MFD from the collector of Q2 to ground. As an aside .001 almost stopped the oscillation, but not quite. My SWAG at the cause of oscillation is leakage caused by too much soldering heat diffusing contaminants into substrate. Some judicious scraping with a dental tool and a bit of conformal coating might also have worked.
If anyone cares the oscillation frequency was 162.7 KHz. With the output filter R29 & C8 removed the frequency increased to 769 KHz about the same amplitude. With C8 value doubled the amplitude (EDIT 2-14-23 Don't recall how the amplitude varied END) and frequency went down to 113 KHz. HP counter and Tek scope.
If you need to replace Q2, Q3 or Q4 on the PA, replace all 3 unless you get an exact 2Sxxx.
During this repair I started reading EchoWars Sticky “Amplifier Distortion, DC-Offset and You”
http://audiokarma.org/forums/showthread.php?t=5634 . I was sorry I had not read or thought about this years ago. I’m up to about page 200 and will continue. If you haven’t read at least the first 3 pages, STOP and do it NOW!
I did use a pair of EchoWars favorite ZTX795A to make a 2SA798 (see pictures) trying to get ride of the 570 mv output offset after curing the oscillations, but that did not help. The final solution to the offset was defluxing the solder side and washing the board with scrubbing bubbles. Removing the .005 cap got the oscillations back after the offset was cured so it is there to stay. The offset now drifts between 2 and 8 mv.
Well that’s about it for now. I did not cover 2 oak cabinets I built for LR9090s, but will detail what I did if it will help anyone. I will be adding additional voltage readings to the LR 9090 schematics and eventually submitting it to the AK database. The RS-660 is a LR-9090 with different bells and whistles.
My friend that was a Lafayette store manager and repair tech told me the LR 120DB schematics did not have much in the way of voltage references as the product was not mature when LR folded. If anyone cares I can add a few and submit it to AK database. That would be from a universe on 1 working unit. Not something I would bet the farm on.